The WHO-UMC Causality Scale

The WHO-UMC causality assessment system, developed and maintained by the Uppsala Monitoring Centre on behalf of the WHO, is the most widely used causality framework in global pharmacovigilance, and it works by asking a structured set of questions rather than asking an assessor to render an unstructured medical opinion. Those questions cluster around four key criteria: is there a plausible time relationship between drug intake and event onset; can the event NOT be reasonably explained by the patient’s underlying disease or other drugs; is the response to withdrawing the drug — the dechallenge — clinically plausible; and, where relevant, was a rechallenge satisfactory, or is the event pharmacologically or phenomenologically definitive on its own.

Those four criteria combine into six categories. Certain requires all of them, including a satisfactory rechallenge — which is exactly why it’s the category assessors reach for least often; deliberately re-exposing a patient to a drug that may have harmed them is rarely done for good ethical reasons, so most cases that would otherwise qualify as Certain simply never get the rechallenge confirmation and land in Probable/Likely instead. Probable/Likely requires a reasonable time relationship and an event unlikely to be explained by disease or other drugs, with a clinically reasonable dechallenge response, but no rechallenge.

Possible sits where a reasonable time relationship exists, but disease or other concomitant drugs could also plausibly explain the event, and dechallenge information may be missing or ambiguous — this is where a large share of real-world spontaneous reports land, because complete information is often simply not available. Unlikely covers cases where the time relationship makes a causal role improbable and another explanation is more plausible. Conditional/Unclassified means the event was reported as an adverse reaction but more data is genuinely needed before a category can be assigned. Unassessable/Unclassifiable is reserved for reports where the available information is insufficient or contradictory and cannot be supplemented or verified — importantly, this isn’t a judgment that the reaction was implausible, only that there isn’t enough usable information to judge it at all.

What makes this scale genuinely useful in practice is that it forces an assessor to work through the same four questions on every case, in the same order, rather than reaching for an overall impression. Two assessors working the same case with the same available information should converge on the same category most of the time — and when they don’t, the disagreement is usually traceable to a specific one of the four criteria being read differently, which is itself useful diagnostic information about where the case is genuinely ambiguous.